IlluminaeAI was a small-scale project I worked on that aimed to provide individualized teaching and feedback to every student in a classroom during lessons and teachings. Similar to how existing systems adapt questions and feedback based on student input, IlluminaeAI took this one step further and added AI to interactively provide and answer questions.
The idea came from talking with my mother, who is a teacher. She told me how hard it is to give every student the one-on-one attention they need when there's one of her and a full classroom of kids, each learning at their own pace. I wanted to build something that could extend her reach, helping students the way she would when she can't be there herself.
However, I want IlluminaeAI to help the teacher, not replace them. As such, our AI models adapt their teaching style to mirror the teachers' using content provided by the teacher and assignments posted in the class's Google Classroom. This feeds the model context like assignments, writing and grading style, etc.
It's tailored to be like your teacher, but available 24/7 for any of your questions.
For the technologies used, we use some cutting-edge RAG technology and frontier models to ensure that the models effectively use the wealth of context available to it.
Finally, to comply with data privacy laws, we check all prompts, context, and responses from the LLM for personally identifable information (PII) and risk factors to ensure the LLM never reads or returns any protected or harmful data.
All of this is made for school districts across the country to provide low-cost, individualized education to all.